Transferable machine learning of excited-state dynamics with extremal pooling
Abstract
Photochemical processes govern phenomena ranging from solar energy conversion and atmospheric chemistry to vision and photosynthesis. Accurate simulation of these processes requires modeling excited-state potential energy surfaces, often involving chemical reactions, tasks that remain computationally prohibitive for extended systems and long timescales using traditional \textit{ab initio} methods. Machine learning interatomic potentials have revolutionized ground-state simulations, but their extension to excited states faces fundamental challenges: standard architectures assume energy extensivity, an assumption that fails for excited states. Here, we present a size-intensive machine-learning framework for excited-state dynamics based on \textit{extremal pooling} of predicted atomic HOMO and LUMO contributions. Trained exclusively on excitations energies and forces, the architecture learns interpretable atomic-level contributions that encode physical information on the extent of electron localization. We demonstrate this framework on the photoexcited solvated electron in liquid water, a paradigmatic problem in radiation chemistry leading to competing pathways involving both hydrogen-atom dissociation and proton-coupled electron transfer. The model not only reproduces the relevant chain of reactions and product species that form during excitation, but also allows one to explicitly study the dynamics of the solvated electron in quantitative agreement with previously reported Restricted Open-Shell Kohn-Sham calculations, while enabling excited-state simulations of periodic systems at length and time scales inaccessible to the reference electronic-structure method. This work establishes a general strategy for machine learning-driven excited-state dynamics applicable to diverse photochemical systems, from molecular chromophores in solution to extended condensed-phase systems.
Related
- Nonadiabatic excited-state dynamics with quantum Monte Carlo-trained machine learning: azomethane as a stringent test
- Electronic configuration vectors provide transferable elemental representations for machine learning across the periodic table
- Efficient and Transferable Machine Learning Potentials for Zn-Coordinated Zeolitic Imidazolate Frameworks Through Structural Diversity Sampling and Transfer Learning.
- Reinforcement Learning Assisted Quantum Simulation of Many-Body Excited States and Real-Time Dynamics
- The Thiolate Shell as a Dynamic Gold Reservoir: Size-Dependent Thermal Restructuring of Gold Nanoclusters from MACE-Based Machine-Learning Molecular Dynamics
BibTeX
@misc{malosso2026transferable,
title = {{Transferable machine learning of excited-state dynamics with extremal pooling}},
author = {Cesare Malosso and Wei Bin How and Gonzalo Díaz Mirón and Ali Hassanali and Michele Ceriotti},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2606.16859},
eprint = {2606.16859},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2606.16859},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Transferable machine learning of excited-state dynamics with extremal pooling
AU - Cesare Malosso
AU - Wei Bin How
AU - Gonzalo Díaz Mirón
AU - Ali Hassanali
AU - Michele Ceriotti
PY - 2026
DA - 2026/10/12
PB - arXiv
DO - 10.48550/arxiv.2606.16859
UR - https://arxiv.org/abs/2606.16859
AB - Photochemical processes govern phenomena ranging from solar energy conversion and atmospheric chemistry to vision and photosynthesis. Accurate simulation of these processes requires modeling excited-state potential energy surfaces, often involving chemical reactions, tasks that remain computationally prohibitive for extended systems and long timescales using traditional \textit{ab initio} methods. Machine learning interatomic potentials have revolutionized ground-state simulations, but their extension to excited states faces fundamental challenges: standard architectures assume energy extensivity, an assumption that fails for excited states. Here, we present a size-intensive machine-learning framework for excited-state dynamics based on \textit{extremal pooling} of predicted atomic HOMO and LUMO contributions. Trained exclusively on excitations energies and forces, the architecture learns interpretable atomic-level contributions that encode physical information on the extent of electron localization. We demonstrate this framework on the photoexcited solvated electron in liquid water, a paradigmatic problem in radiation chemistry leading to competing pathways involving both hydrogen-atom dissociation and proton-coupled electron transfer. The model not only reproduces the relevant chain of reactions and product species that form during excitation, but also allows one to explicitly study the dynamics of the solvated electron in quantitative agreement with previously reported Restricted Open-Shell Kohn-Sham calculations, while enabling excited-state simulations of periodic systems at length and time scales inaccessible to the reference electronic-structure method. This work establishes a general strategy for machine learning-driven excited-state dynamics applicable to diverse photochemical systems, from molecular chromophores in solution to extended condensed-phase systems.
ER -